In [5]:
np.random.seed(1)
plt.figure(figsize = (15,20))
for i in range(1,17):
image = X[i]
label = y[i]
plt.subplot(4,4,i)
plt.imshow(image)
plt.title(label)
plt.show()
### import libraries
import os
### Core
import numpy as np
from skimage.transform import resize
from sklearn.model_selection import train_test_split
from sklearn.utils import shuffle
### Deep Learning
import keras
from keras.models import Sequential
from keras.layers import Dense, Flatten, Conv2D, MaxPooling2D, Dropout
from keras import callbacks
from tensorflow.keras.utils import to_categorical
### Visualization
import matplotlib.pyplot as plt
import plotly.express as px
%%time
categories = {"vehicles":1, "non-vehicles":0}
images = []
target = []
# cycle through labels
for label in os.listdir("data"):
# get label directory
imagedir = os.path.join("data", label)
label_value = categories[label]
print(f"label: {label}, value: {label_value}, imagedir: {imagedir}")
# get images from each label directory
for image in os.listdir(imagedir):
imagepath = os.path.join(imagedir, image)
img = plt.imread(imagepath)
resized_img = resize(img, (64,64,3))
images.append(resized_img)
target.append(label_value)
label: non-vehicles, value: 0, imagedir: data\non-vehicles label: vehicles, value: 1, imagedir: data\vehicles CPU times: total: 5.17 s Wall time: 1min 56s
n_classes = 2
classes = ["Vehicles","Non-Vehicles"]
class_distribution = [sum(target), len(target) - sum(target)]
plt.figure(figsize = (8,6))
fig = px.pie(names = classes,
values = class_distribution,
title = "Class Distribution",
color_discrete_sequence = px.colors.qualitative.Vivid
)
fig.update_layout({'title':{'x':0.5}})
# fig.write_image('../images/piechart_class.png', format = 'png') # takes forever to write -- manually save image
fig.show()
<Figure size 576x432 with 0 Axes>
X = np.array(images)
y = np.array(target)
X, y = shuffle(X, y, random_state=0)
np.random.seed(1)
plt.figure(figsize = (15,20))
for i in range(1,17):
image = X[i]
label = y[i]
plt.subplot(4,4,i)
plt.imshow(image)
plt.title(label)
plt.show()
# Split Data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)
y_train_one_hot = to_categorical(y_train)
y_test_one_hot = to_categorical(y_test)
print(f"Shape X_train: {X_train.shape}")
print(f"Shape y_train: {y_train.shape}")
print(f"Shape X_test: {X_test.shape}")
print(f"Shape y_test: {y_test.shape}")
Shape X_train: (14208, 64, 64, 3) Shape y_train: (14208,) Shape X_test: (3552, 64, 64, 3) Shape y_test: (3552,)
model_name = "Vehicle_Detection"
# Begin Model Sequence
model = Sequential()
model.add(Conv2D(filters = 32, kernel_size = (5,5), activation='relu', input_shape = X_train[0].shape))
model.add(MaxPooling2D(pool_size = (2,2)))
model.add(Conv2D(filters = 32, kernel_size = (5,5), activation='relu'))
model.add(MaxPooling2D(pool_size = (2,2)))
# Flatten
model.add(Flatten())
# Dense layers
model.add(Dense(32,activation = 'relu'))
model.add(Dropout(0.1))
model.add(Dense(64,activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(2,activation = 'softmax'))
# Compile Model
model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])
# Callbacks
CB = [
callbacks.EarlyStopping(patience=5, restore_best_weights=True),
callbacks.ModelCheckpoint(model_name + ".h5", save_best_only=True)
]
history = model.fit(X_train,y_train_one_hot, epochs = 50, callbacks = CB, validation_split = 0.2)
Epoch 1/50 356/356 [==============================] - 22s 61ms/step - loss: 0.2615 - accuracy: 0.8948 - val_loss: 0.1257 - val_accuracy: 0.9514 Epoch 2/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0952 - accuracy: 0.9684 - val_loss: 0.0602 - val_accuracy: 0.9792 Epoch 3/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0545 - accuracy: 0.9819 - val_loss: 0.1147 - val_accuracy: 0.9536 Epoch 4/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0428 - accuracy: 0.9858 - val_loss: 0.0506 - val_accuracy: 0.9810 Epoch 5/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0401 - accuracy: 0.9871 - val_loss: 0.0806 - val_accuracy: 0.9669 Epoch 6/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0469 - accuracy: 0.9833 - val_loss: 0.0383 - val_accuracy: 0.9877 Epoch 7/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0273 - accuracy: 0.9911 - val_loss: 0.0653 - val_accuracy: 0.9799 Epoch 8/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0319 - accuracy: 0.9891 - val_loss: 0.0319 - val_accuracy: 0.9884 Epoch 9/50 356/356 [==============================] - 22s 63ms/step - loss: 0.0309 - accuracy: 0.9903 - val_loss: 0.0336 - val_accuracy: 0.9880 Epoch 10/50 356/356 [==============================] - 22s 63ms/step - loss: 0.0159 - accuracy: 0.9950 - val_loss: 0.0398 - val_accuracy: 0.9884 Epoch 11/50 356/356 [==============================] - 22s 63ms/step - loss: 0.0138 - accuracy: 0.9952 - val_loss: 0.0249 - val_accuracy: 0.9919 Epoch 12/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0172 - accuracy: 0.9942 - val_loss: 0.0414 - val_accuracy: 0.9891 Epoch 13/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0184 - accuracy: 0.9941 - val_loss: 0.0627 - val_accuracy: 0.9761 Epoch 14/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0179 - accuracy: 0.9938 - val_loss: 0.0322 - val_accuracy: 0.9884 Epoch 15/50 356/356 [==============================] - 22s 63ms/step - loss: 0.0083 - accuracy: 0.9972 - val_loss: 0.0216 - val_accuracy: 0.9933 Epoch 16/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0151 - accuracy: 0.9950 - val_loss: 0.0230 - val_accuracy: 0.9919 Epoch 17/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0159 - accuracy: 0.9952 - val_loss: 0.0165 - val_accuracy: 0.9954 Epoch 18/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0158 - accuracy: 0.9945 - val_loss: 0.0192 - val_accuracy: 0.9937 Epoch 19/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0042 - accuracy: 0.9992 - val_loss: 0.0477 - val_accuracy: 0.9842 Epoch 20/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0140 - accuracy: 0.9949 - val_loss: 0.0289 - val_accuracy: 0.9891 Epoch 21/50 356/356 [==============================] - 22s 62ms/step - loss: 0.0054 - accuracy: 0.9982 - val_loss: 0.0247 - val_accuracy: 0.9937 Epoch 22/50 356/356 [==============================] - 22s 61ms/step - loss: 0.0089 - accuracy: 0.9976 - val_loss: 0.0284 - val_accuracy: 0.9926
accuracy = history.history["accuracy"]
val_accuracy = history.history["val_accuracy"]
plt.plot(accuracy)
plt.plot(val_accuracy)
plt.title("Model Accuracy")
plt.ylabel("Accuracy")
plt.xlabel('Epochs')
plt.legend(['Train','Val'])
plt.show()
loss = history.history["loss"]
val_loss = history.history["val_loss"]
plt.plot(loss)
plt.plot(val_loss)
plt.title("Model Accuracy")
plt.ylabel("Loss")
plt.xlabel('Epochs')
plt.legend(['Train','Val'])
plt.show()
model = keras.models.load_model("Vehicle_Detection.h5")
test_accuracy = model.evaluate(X_test,y_test_one_hot)[1]
test_accuracy = '%.3f'%test_accuracy
print(f"Testing Accuracy: {test_accuracy}")
111/111 [==============================] - 1s 11ms/step - loss: 0.0249 - accuracy: 0.9935 Testing Accuracy: 0.994